New method enhances covariate selection for causal inference in machine learning

A new research paper introduces a method for improving covariate selection in doubly robust double/debiased machine learning (DML) for causal inference. The proposed approach involves using the union of covariates selected by both the propensity score and outcome models to re-estimate these models. This technique aims to reduce confounding bias more effectively than using separate covariate sets, as demonstrated by simulation results. The findings also suggest that machine learning-based estimation does not always outperform traditional doubly robust estimation, and that post-Lasso methods can reduce more confounding bias than standard Lasso. AI

IMPACT Enhances the accuracy of causal inference models by improving covariate selection, potentially leading to more reliable insights from complex datasets.

RANK_REASON The cluster contains a research paper detailing a new methodology for machine learning in causal inference. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv stat.ML →

• causal inference
• Double Robustness
• Doubly Robust Double/debiased Machine Learning
• lasso
• machine learning
• Post-Lasso
• propensity score matching

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →